{
  "as_of": "2026-09-29",
  "resource_baseline": "9f0f633f",
  "language_choices": 8,
  "total_pairs": 678843,
  "total_reference_lines": 979,
  "languages": [
    {
      "language": "Hindi / Hinglish",
      "code": "hin",
      "pairs": 130284,
      "roman_spellings": 50000,
      "reference_lines": 340,
      "resource_sha256": "c1291bcef0a095b0886572a417a1524f63eb13d49ef1eae15a49fd66f064e462",
      "fixture_sha256": {
        "golden-hinglish.tsv": "697b843179efdc95bdfaf4cae81d102476264718153f9a2275dc0ebe82a1d13c"
      }
    },
    {
      "language": "Telugu",
      "code": "tel",
      "pairs": 114073,
      "roman_spellings": 50000,
      "reference_lines": 123,
      "resource_sha256": "ddabe03923905d46ed6ac75aa2d41d0660e64ff013662ca7790a9e8a15905457",
      "fixture_sha256": {
        "golden-telugu.tsv": "d1185813b4e8e9b8f00c4a4b6b611f1b52d9fbb729e13fd82a3b63dd54f60cb3"
      }
    },
    {
      "language": "Tamil",
      "code": "tam",
      "pairs": 106473,
      "roman_spellings": 50000,
      "reference_lines": 118,
      "resource_sha256": "4c07af319072032c8a4e3d6d99224ccf7d4589c77a62809b8eb76638a3a28e5b",
      "fixture_sha256": {
        "golden-tamil.tsv": "aa90e497ed2abda668221947ddaed8ac3b2fa320b0361dc8b73f513dfcff03ec"
      }
    },
    {
      "language": "Kannada",
      "code": "kan",
      "pairs": 115975,
      "roman_spellings": 50000,
      "reference_lines": 118,
      "resource_sha256": "c8930d17613ac2190246cb2a89a10c59222fb4b70c328637224c453220959627",
      "fixture_sha256": {
        "golden-kannada.tsv": "1656d8ae5c10568e950ca06a8b846f112d46e2d9126fa56a2ced982130ccb3b1"
      }
    },
    {
      "language": "Marathi",
      "code": "mar",
      "pairs": 110129,
      "roman_spellings": 50000,
      "reference_lines": 119,
      "resource_sha256": "1d3f19f472ad2b5d8a2df12a8674cfbced276df1a4c92279bcc537df953391e4",
      "fixture_sha256": {
        "golden-marathi.tsv": "aac40a12af4536a889dd106b088e3cd721d9c19c308b259844faf3ece7968beb"
      }
    },
    {
      "language": "Punjabi",
      "code": "pan",
      "pairs": 101909,
      "roman_spellings": 50000,
      "reference_lines": 161,
      "resource_sha256": "53baadd3114da55820edea1d937331508e3188204595031378b70eacb352e2a2",
      "fixture_sha256": {
        "golden-punjabi.tsv": "2353bf06cf8c3996a4b37b7dff7efdf626a819774e006bb59247f0a25a56f487"
      }
    }
  ],
  "methodology": "Bundled native-language dictionaries only. Count distinct (lowercase Roman spelling, native form) pairs after decompression, excluding comments and invalid rows. Count two-column reference-line fixtures excluding comments and blank lines. Hindi and Hinglish share a dictionary; English is excluded from pair totals. Counts describe resource size and test-set size, not model parameters, training examples, test pass rates, audio accuracy or competitor performance.",
  "evaluation_status": "Internal text fixtures. Telugu, Tamil, Kannada, Marathi and Punjabi sets await native-speaker review. No controlled public cross-provider audio benchmark has been completed.",
  "data_sources": [
    {
      "name": "AI4Bharat Aksharantar",
      "url": "https://huggingface.co/datasets/ai4bharat/Aksharantar"
    },
    {
      "name": "SCOWL English word lists",
      "url": "https://wordlist.aspell.net/"
    }
  ]
}
